About the role#
The Time Series Research and Development team at SAS focuses on advancing forecasting and artificial intelligence capabilities for temporal data. As an intern, you will work on researching modern machine-learning methods, developing scalable production software, and integrating new technologies with the SAS Viya platform. You will contribute to real projects that impact our forecasting tools.
What you'll need#
- Currently enrolled as a master's-level student in Statistics, Econometrics, Computer Science, Machine Learning, Artificial Intelligence, or a related quantitative field, with a graduation date no earlier than December 2027.
- Experience with deep learning and transformer architectures, including hands-on model development in Python using PyTorch, TensorFlow, or a comparable framework.
- Knowledge of time series analysis, forecasting, representation learning, or generative modeling.
- Experience preparing large datasets and designing rigorous machine-learning experiments and evaluation pipelines.
- Ability to read research papers, translate ideas into working prototypes, and communicate technical results clearly.
- Strong written and verbal communication skills and demonstrated leadership through initiative.
- Interest in the future of Analytics.
- Preferred: Experience with time series foundation models, large-scale pretraining, zero-shot or few-shot learning, generative models, distributed or multi-GPU training, C or C++, and publications in relevant venues.
- Must be legally authorized to work in the United States without requiring sponsorship now or in the future.
What you'll do#
- Research and compare open-source and commercial time series foundation models, focusing on architecture, pretraining data, training objectives, adaptation strategies, and evaluation methodology.
- Curate, clean, standardize, and document time series datasets for large-scale pretraining and evaluation.
- Design and implement components of a transformer-based time series foundation model for SAS forecasting use cases.
- Develop and run distributed pretraining experiments, analyze model behavior, and improve zero-shot and few-shot forecasting performance.
- Build reproducible benchmarks against open-source foundation models and SAS forecasting intellectual property using accuracy, robustness, efficiency, and generalization criteria.
- Explore transformer-based synthetic time series generation as pretraining data and as a standalone capability.
- Prototype integration with SAS Viya and communicate research findings through technical reports, code, and presentations.
Location & details#
- Location: Cary, North Carolina (Hybrid).
- Availability: Must be local to Cary HQ and available to start in person on May 18, 2027.
- Term: Summer 2027 (May 18, 2027 - August 6, 2027).
- This is a paid, full-time internship.


